Volcano deformation monitoring is fundamental to detect pressurizations of magma bodies and forecasting any ensuing eruptions. Analytical and quasi-analytical solutions for pressurized cavities are routinely used to constrain volcano deformation sources through inversion of surface displacement data. Due to their computational efficiency, such solutions enable a thorough exploration of the parameter space and thereby provide insight into the physics of magma-rock interaction. Developing more general deformation models can help us better characterize subsurface magma storage. We develop quasi-analytical solutions for the surface deformation field due to the pressurization of a finite (triaxial) ellipsoidal cavity in a half-space. The solution is in the form of a non-uniform distribution of triaxial point sources within the cavity. The point sources have the same aspect ratio, determined by the cavity shape, while their strengths and spacing are determined in an adaptive manner, such that the net point-source potency per unit volume is uniform. We validate and compare our solution with analytical and numerical solutions. We provide computationally-efficient MATLAB codes tailored for source inversions. This solution opens the possibility of exploring the geometry of shallow magma chambers for potential deviations from axial symmetry.

GPS observations of ocean tide loading displacements can help infer the regional anelastic properties of the asthenosphere. We estimate M2 ocean tide loading displacements at 170 GPS sites in New Zealand and compare these to modeled values using a range of numerical tide and radially symmetric (1D) elastic and anelastic Earth models. Regardless of the model combination we are unable to reduce the strong spatial coherence of the M2 residuals across the North Island where they reach 0.4 mm (2%). The best fit in the North Island is obtained when combining the FES2014b tide model with spatially-variable ocean density and water compressibility, and the STW105 Earth model. The residuals exhibit a change of ~0.3 mm in magnitude between the Taupo Volcanic Zone and the east coast (~100 km), suggesting that this region’s laterally-varying, shallow rheological structure may need to be considered to explain these observations.

Gravity fluctuations produced by ambient seismic fields are predicted to limit the sensitivity of the next-generation, gravitational-wave detector Einstein Telescope at frequencies below 20 Hz. The detector will be hosted in an underground infrastructure to reduce seismic disturbances and associated gravity fluctuations. Additional mitigation might be required by monitoring the seismic field and using the data to estimate the associated gravity fluctuations and to subtract the estimate from the detector data, a technique called coherent noise cancellation. In this paper, we present a calculation of correlations between surface displacement of a seismic field and the associated gravitational fluctuations using the spectral-element SPECFEM3D Cartesian software. The model takes into account the local topography at a candidate site of the Einstein Telescope at Sardinia. This paper is a first demonstration of SPECFEM3D's capabilities to provide estimates of gravitoelastic correlations, which are required for an optimized deployment of seismometers for gravity-noise cancellation.

The forecasting of local GIC effects has largely relied on the forecasting of dB/dt as a proxy and, to date, little attention has been paid to directly forecasting the geoelectric field or GICs themselves. We approach this problem with machine learning tools, specifically recurrent neural networks or LSTMs by taking solar wind observations as input and training the models to predict two different kinds of output: first, the geoelectric field components Ex and Ey; and second, the GICs in specific substations in Austria. The training is carried out on the geoelectric field and GICs modelled from 26 years of one-minute geomagnetic field measurements, and results are compared to GIC measurements from recent years. The GICs are generally predicted better by an LSTM trained on values from a specific substation, but only a fraction of the largest GICs are correctly predicted. This model had a correlation with measurements of around 0.6, and a root-mean-square error of 0.7 A. The probability of detecting mild activity in GICs is around 50%, and 15% for larger GICs.

The continuous estimation of changes in seismic velocity and seismic scattering property by passive interferometry using seismic ambient noise is a promising tool for monitoring volcanoes. To improve the usefulness of this method, it is necessary not only to detect subsurface structural changes but also to quantitatively compare the estimated changes in seismic wave velocity and seismic wave scattering property with other observations such as ground deformation. We applied passive interferometry to continuous seismic records from Suwanosejima volcano, Japan, recorded between April 2017 and December 2021. We detected repeated significant waveform decorrelations in seismic ambient noise cross-correlation functions, indicating seismic scattering property changes in the shallow areas of the volcano. These decorrelations were observed from 2 week to a few days before the increase in the number of explosions, suggesting that seismic scattering properties changed significantly during that period. We found that the timing of the decorrelation in seismic ambient noise cross-correlation functions and tilt changes related to magma accumulation and injection beneath Suwanosejima were well synchronized. The high correlation between the amounts of decorrelation and tilt change during the magma accumulation period suggests that a large volume of accumulated magma caused great changes in the scattering property. These results provide a significant first step toward a quantitative comparison of the amount of changes in the scattering property with the amount of magma accumulation beneath volcanoes.

Deformation bands are the main structural element of fault damage zones within sandstone reservoirs. The prediction of band occurrence and their petrophysical impacts is based largely on the understanding that the yield and deformation mechanism of sandstones is primarily controlled by porosity and mean grain size. Whilst this is supported by field observations within aeolian successions, where bands are predictably favoured within coarse-grained, high-porosity sandstones, the prediction of deformation bands within texturally complex mixed aeolian-fluvial reservoirs on the basis of porosity and grain size alone, may be unreliable. The effect of grain sorting on the mechanical behaviour of sandstones is not well understood, although it is generally regarded that deformation band formation is inhibited in texturally immature sandstones with a poor level of sorting. We examine the effect of sorting on both the inelastic yield of sandstones, the dominant deformation mechanism by which yield occurs, and the textural and microstructural changes with deformation, using a series of triaxial experiments on unconsolidated quartz sands. Hydrostatic experiments were conducted on over-consolidated samples of very well- to moderately-sorted sands with a range of mean grain sizes from 128-700µm. We report accurate prediction of P* using porosity x grain radius, with P* reduced with decreased sorting. Constant displacement rate triaxial experiments are performed at up to 10% axial strain to explore yield behaviour in both the brittle dilatant regime and shear-enhanced compactive regime. Experiments were repeated with systematically varied grain sorting whilst mean grain size and porosity was maintained. The textural and petrophysical changes are observed and quantified using pore volumometry, back scattered electron microscopy, digital image analysis and point counting. Results show that in well-sorted sands, localised cataclasis and deformation band formation is the dominant deformation mechanism. In poorly-sorted sands deformation occurs through a combination of grain boundary sliding and randomly distributed pockets of cataclasis. Using grain size analysis we identify greater levels of cataclasis and production of fines in well-sorted sands, resulting in permeability reduction up to one order of magnitude more than that of poorly-sorted sands deformed at the same conditions. We hypothesise that band formation within poorly sorted sandstones may be promoted by the formation and propagation of bands in adjacent well sorted sandstones where band formation is favoured. These results give insight into the deformation, textural changes, and permeability impact of both unconsolidated and consolidated siliciclastic reservoirs.

New geophysical data from Antarctica’s Ross Embayment reveal the structure and subglacial geology of extended continental crust beneath the Ross Ice Shelf. We use airborne magnetic data from the ROSETTA-Ice Project to locate the contact between magnetic basement and overlying sediments. We delineate a broad, segmented basement high with thin (0-500m) non-magnetic sedimentary cover which trends northward into the Ross Sea’s Central High. Before subsiding in the Oligocene, this feature likely facilitated early glaciation in the region and subsequently acted as a pinning point and ice flow divide. Flanking the high are wide sedimentary basins, up to 3700m deep, which parallel the Ross Sea basins and likely formed during Cretaceous-Neogene intracontinental extension. NW-SE basins beneath the Siple Coast grounding zone, by contrast, are narrow, deep, and elongate. They suggest tectonic divergence upon active faults that may localize geothermal heat and/or groundwater flow, both important components of the subglacial system.

We investigate the flow-wise variation of the hydraulic conductivity inside a non-uniformly shaped fracture with permeable walls. Using lubrication theory for viscous flows, in conjunction with the Beavers–Joseph–Saffman boundary condition at the permeable walls, we obtain an analytical expression for the velocity profile, conductivity, and wall permeation velocity. These predictions highlight the effects of geometric variation (through the local slope of the aperture’s flow-wise variation), the permeability of the walls (through a dimensionless slip coefficient), and the effect of flow inertia (through a Reynolds number). The theory is validated against an OpenFOAM(R) solver for the Navier–Stokes equations subject to a tensorial slip boundary condition, showing good agreement. The mathematical results have implications on system-level (multiscale) modeling of hydraulically fractured reservoirs, in which the Darcy conductivity of each non-uniform passage must be accurately accounted for, throughout the fractured porous rock.

Earth’s internal heat drives its dynamic engine, causing mantle convection, plate tectonics, and the geodynamo. These renewing and protective processes, which make Earth habitable, are fueled by a primordial (kinetic) and radiogenic heat. For the past two decades, particle physicists have measured the flux of geoneutrinos, electron antineutrinos emitted during β − decay. These ghost-like particles provide a direct measure of the amount of heat producing elements (HPE: Th & U) in the Earth and in turn define the planet’s absolute concentration of the refractory elements. The geoneutrino flux has contributions from the lithosphere and mantle. Detector sensitivity follows a 1/r 2 (source detector separation distance) dependence. Accordingly, an accurate geologic model of the Near-Field Lithosphere (NFL, closest 500 km) surrounding each experiment is required to define the mantle’s contribution. Because of its proximity to the detector and enrichment in HPEs, the local lithosphere contributes ∼50% of the signal and has the greatest effect on interpreting the mantle’s signal. We re-analyzed the upper crustal compositional model used by Agostini et al. (2020) for the Borexino experiment. We documented the geology of the western Near-Field region as rich in potassic volcanism, including some centers within 50 km of the detector. In contrast, the Agostini study did not include these lithologies and used only a HPE-poor, carbonate-rich, model for upper crustal rocks in the surrounding ∼150 km of the Borexino experiment. Consequently, we report 3× higher U content for the local upper crust, which produces a 200% decrease in Earth’s radiogenic heat budget, when compared to their study. Results from the KamLAND and Borexino geoneutrino experiments are at odds with one another and predict mantle compositional heterogeneity that is untenable. Combined analyses of the KamLAND and Borexino experiments using our revised local models strongly favor an Earth with ∼20 TW present-day total radiogenic power. The next generation of geoneutrino detectors (SNO+, counting; and JUNO, under construction) will better constrain the HPE budget of the Earth.

The discipline of land change science has been evolving rapidly in the past decades. Remote sensing played a major role in one of the essential components of land change science, which includes observation, monitoring, and characterization of land change. In this paper, we proposed a new framework of the multifaceted view of land change through the lens of remote sensing and recommended five facets of land change including change location, time, target, metric, and agent. We also evaluated the impacts of spatial, spectral, temporal, angular, and data-integration domains of the remotely sensed data on observing, monitoring, and characterization of different facets of land change, as well as discussed some of the current land change products. We recommend clarifying the specific land change facet being studied in remote sensing of land change, reporting multiple or all facets of land change in remote sensing products, shifting the focus from land cover change to specific change metric and agent, integrating social science data and multi-sensor datasets for a deeper and fuller understanding of land change, and recognizing limitations and weaknesses of remote sensing in land change studies.

Advances in space weather science and small satellite (SmallSat) technology have proceeded in parallel over the past two decades, but better communication and coordination is needed among the respective worldwide communities contributing to this rapid progress. We identify six areas where improved international coordination is especially desirable, including: (1) orbital debris miti-gation; (2) spectrum management; (3) export control regulations; (4) access to timely and low-cost launch opportunities; (5) inclusive data policies; and (6) education. We argue the need for interna-tionally coordinated policies and programs to promote the use of SmallSats for space weather re-search and forecasting while realizing maximum scientific and technical advances through the inte-gration of these two increasingly important endeavors.

Tidal triggering of tectonic tremors has been observed at plate boundaries around the circum-Pacific region. It has been reported that the response of tremors to tidal stress during episodic tremor and slow slip (ETS) changes between the early and later stages of ETS. Several physical models have been constructed, with which observations for the tidal response during ETS have been partly reproduced. However, no model has been proposed that reproduces all the observations. In this study, a model adopted in previous studies is extended to include the effects of dilatancy/compaction that occur in the fault creep region. The analytical approximate solution derived in this study and numerical computational results reveal how the tidal response depends on the physical properties of the fault. Furthermore, the model reproduces all the above observations simultaneously for a specific range of fault parameters. Of particular importance is that the occurrence of dilatancy/compaction is essential to reproduce the tidal response at the early stage of the ETS. The value of the critical distance dc is constrained to be approximately　1~10 cm. This agrees with the values that have been widely used in seismic cycle numerical simulations rather than those obtained in laboratory experiments. The fluid pressure diffusivity is constrained to be at least 10^(-5) m^2/s or less, and the effective normal stress is constrained to 10^(5~6) Pa. In conclusion, this study shows that reproducing the tidal response of tectonic tremors during the ETS is useful for estimating fault physical properties, including hydraulic properties.

This article introduces factors contributing significantly to climate change that have been largely neglected in both the scientific and popular press. These factors have immediate implications for public policy directed at slowing, halting and even reversing climate change and its effects. This article argues that in addition to the known contributions made by greenhouse gasses, climate change is also driven by shifts in the patterns of global atmospheric circulation which are influenced by persistent, large-scale vortices caused by the wake turbulence left by commercial air traffic. Because this traffic is highly concentrated along the most frequently traveled routes, the vortices aircraft create have transformed into semi-permanent atmospheric circulation which have widespread effects on how the atmosphere traps and releases heat. It is also possible that these changes alter the loss of water from the atmosphere. This would endanger all life on earth, not just the human population.

Stress accumulation and release in the crust remains poorly understood compared to that at the plate boundaries. Spatiotemporal variations in foreshock and aftershock activities can provide key constraints on time-dependent stress and deformation processes in the crust. The 2017 M5.2 Akita-Daisen intraplate earthquake in NE Japan was preceded by intense foreshock activity and triggered a strong sequence of aftershocks. We examine the spatiotemporal distributions of foreshocks and aftershocks and determine the coseismic slip distribution of the mainshock. Our results indicate that seismicity both before and after the mainshock was concentrated on a planar structure with N-S strike that dips steeply eastward. We observe a migration of foreshocks towards the mainshock rupture area, suggesting that foreshocks were triggered by aseismic phenomena preceding the mainshock. The mainshock rupture propagated toward the north, showing less slip beneath foreshock regions. The stress drop of the mainshock was 1.4 MPa and the radiation efficiency was 0.72. Aftershocks were intensely triggered near the edge of large coseismic slip regions where shear stress increased. The aftershock region expanded along the fault strike, which is attributed to the post-seismic aseismic slip of the mainshock. The postseismic slip possibly triggered repeating earthquakes with M ~3. We find that the foreshocks, mainshock, aftershocks, and post-seismic slip released stress at different segments along the fault, which may reflect differences in frictional properties. Obtained results were similar to those observed for interplate earthquakes, which supports the hypothesis that the deformation processes along plate boundaries and crustal faults are fundamentally the same.

As a result of uneven density of data collection, level-2 satellite gravimetry data suffer from global north-south striping. By applying various filtering methods, several studies have addressed the mitigation of the data. However, the studies mainly addressed the issue on a global scale, and the local effects were not considered. On the other hand, water research, especially inland hydrology, usually deals with small-scale fitures such as lakes and watersheds. Therefore, the local data de-striping methods need special attention. This research presents a new analytical method to de-stripe gravimetry data based on the spatial contrast of signals. The approach strikes a balance between de-striping and signal preservation. Using a-priori information obtained from the gravimetry data, the de-striping method first estimates the spatial gradient of the signal and optimizes a Poisson filter based on this information to de-stripe the data. Unlike the other approaches, the optimized filter is dynamic and accounts for temporal variations in the signal contrast, such as seasonality. The proposed approach is applied to ten globally distributed study areas to derive a general scheme. Detailed processes and evaluations are applied to two study areas: the Caspian Sea and the Congo River Basin. Results are visually assessed for spatial fit and for temporal consistency by comparison with results from other filters. The use of a dynamic filter set specified for each region and time point allows us to preserve local hydrologic signals that are susceptible to globally optimized filters. It also allows filter-related errors to be effectively constrained.

Transient deformations associated with foreshocks activity has been observed before large earthquakes, suggesting the occurrence of a detectable pre-seismic slow slip during the initiation phase. In this respect, a critical issue consists in discriminating the relative contributions from seismic and aseismic fault slip during the preparation phase of large earthquakes. We focus on the April-May 2017 Valparaíso earthquake sequence, which involved a Mw=6.9 earthquake preceded by an intense foreshock activity. To assess the relative contribution of seismic and aseismic slip, we compare surface displacements predicted from foreshock source models to the transient motion measured prior to the mainshock. The comparison between observed and predicted displacements shows that only half of the total displacement can be explained by the contribution of foreshocks. This result suggests the presence of aseismic preslip during an initiation phase preceding the mainshock.

The instrument package SEIS (Seismic Experiment for Internal Structure) with the three very broadband and three short-period seismic sensors is installed on the surface on Mars as part of NASA’s InSight Discovery mission. When compared to terrestrial installations, SEIS is deployed in a very harsh wind and temperature environment that leads to inevitable degradation of the quality of the recorded data. One ubiquitous artifact in the raw data is an abundance of transient one-sided pulses often accompanied by high-frequency spikes. These pulses, which we term “glitches”, can be modeled as the response of the instrument to a step in acceleration, while the spikes can be modeled as the response to a simultaneous step in displacement. We attribute the glitches primarily to SEIS-internal stress relaxations caused by the large temperature variations to which the instrument is exposed during a Martian day. Only a small fraction of glitches correspond to a motion of the SEIS package as a whole caused by minuscule tilts of either the instrument or the ground. In this study, we focus on the analysis of the glitch+spike phenomenon and present how these signals can be automatically detected and removed from SEIS’ raw data. As glitches affect many standard seismological analysis methods such as receiver functions, spectral decomposition and source inversions, we anticipate that studies of the Martian seismicity as well as studies of Mars’ internal structure should benefit from deglitched seismic data.

Explosions and earthquakes are effectively discriminated by P/S amplitude ratios for moderate magnitude events (M≥4) observed at regional to teleseismic distances (≥200 km). It is less clear if P/S ratios are effective explosion discriminants for lower magnitudes observed at shorter distances. We report new tests of P/S discrimination using a dense seismic array in a continental volcanic arc setting near Mount St. Helens, with 23 single-fired borehole explosions (ML 0.9-2.3) and 406 earthquakes (ML 1-3.3). The array provides up to 95 three-component broadband seismographs and most source-receiver distances are <120 km. Additional insight is provided by ~3,000 vertical component geophone recordings of each explosion. Potential controls on local distance P/S ratios are investigated, including: frequency range, distance, magnitude, source depth, number of seismographs, and site effects. A frequency band of about 10-18 Hz performs better than lower or narrower bands because explosion-induced S-wave amplitudes diminish relative to P for higher frequencies. Source depth and magnitude exhibited weak influences on P/S ratios. Site responses for earthquakes and explosions are correlated with each other and with shallow crustal Vp and Vs from travel-time tomography. Overall, the results indicate high potential for local distance P/S explosion discrimination in a continental volcanic arc setting, with ≥98% true positives and ≤6.3% false positives when using the array median from ≥16 stations. Performance is reduced for smaller arrays, especially those with ≤4 stations, thereby emphasizing the importance of array data for discrimination of low magnitude explosions.

Twenty-two sites, subjected to an IZZI-modified Thellier-Thellier experiment and strict selection criteria, recover a paleomagnetic axial dipole moment (PADM) of 62.24$\pm$ 30.6 ZAm$^2$ in Northern Israel over the Pleistocene (0.012 - 2.58 Ma). Pleistocene data from comparable studies from Antarctica, Iceland, and Hawaii, re-analyzed using the same criteria and age range, show that the Northern Israeli data are on average slightly higher than those from Iceland (PADM = 53.8 $\pm$ 23 ZAm$^2$, n = 51 sites) and even higher than the Antarctica average %\cite{asefaw21} (PADM = 40.3 $\pm$ 17.3 ZAm$^2$, n = 42 sites). Also, the data from the Hawaiian drill core, HSDP2, spanning the last half million years (PADM = 76.7 $\pm$ 21.3 ZAm$^2$, n = 59 sites) are higher than those from Northern Israel. These results, when compared to Pleistocene results filtered from the PINT database (www.pintdb.org) suggest that data from the Northern hemisphere mid-latitudes are on average higher than those from the southern hemisphere and than those from latitudes higher than 60$^{\circ}$N. The weaker intensities found at high (northern and southern) latitudes therefore, cannot be attributed to inadequate spatio-temporal sampling of a time-varying dipole moment or low quality data. The high fields in mid-latitude Northern hemisphere could result from long-lived non-axial dipole terms in the geomagnetic field with episodes of high field intensities occurring at different times in different longitudes. This hypothesis is supported by an asymmetry predicted from the Holocene, 100 kyr, and five million year time-averaged geomagnetic field models.

Geomagnetically induced currents (GICs) at middle latitudes have received increased attention after reported power-grid disruptions due to geomagnetic disturbances. However, quantifying the risk to the electric power grid at middle latitudes is difficult without understanding how the GIC sensors respond to geomagnetic activity on a daily basis. Therefore, in this study the question “Do measured GICs have distinguishable and quantifiable long- and short-period characteristics?” is addressed. The study focuses on the long-term variability of measured GIC, and establishes the extent to which the variability relates to quiet-time geomagnetic activity. GIC quiet-day curves (QDCs) are computed from measured data for each GIC node, covering all four seasons, and then compared with the seasonal variability of Thermosphere-Ionosphere- Electrodynamics General Circulation Model (TIE-GCM)-simulated neutral wind and height-integrated current density. The results show strong evidence that the middle-latitude nodes routinely respond to the tidal-driven Sq variation, with a local time and seasonal dependence on the the direction of the ionospheric currents, which is specific to each node. The strong dependence of GICs on the Sq currents demonstrates that the GIC QDCs may be employed as a robust baseline from which to quantify the significance of GICs during geomagnetically active times and to isolate those variations to study independently. The QDC-based significance score computed in this study provides power utilities with a node-specific measure of the geomagnetic significance of a given GIC observation. Finally, this study shows that the power grid acts as a giant sensor that may detect ionospheric current systems.

Our understanding of geomagnetic field intensity prior to the era of direct instrumental measurements relies on paleointensity analysis of rocks and archaeological materials that serve as magnetic recorders. Only in rare cases absolute paleointensity datasets are continuous over millennial timescales, in sub-centennial resolution, and directly dated using radiocarbon. As a result, fundamental properties of the geomagnetic field, such as its maximal intensity and change rate have remained a subject of lively discussion. Here, we place firm constraints on these two quantities using Bayesian modeling of well-dated archaeomagnetic intensity data from the Levant and Upper Mesopotamia. We report new data from 23 groups of pottery collected from 18 consecutive radiocarbon-dated archaeological strata from Tel Megiddo, Israel. In the Near East, the period of 1700–550 BCE is represented by 84 groups of archaeological artifacts, 55 of which were dated using radiocarbon or a direct link to clear historically-dated events, providing unprecedented sub-century resolution. Moreover, stratigraphic relationships between samples collected from multi-layered sites enable further refinement of the data ages. The Bayesian curve shows four geomagnetic spikes between 1050 and 600 BCE, with virtual axial dipole moment (VADM) reaching values of 155–162 ZAm2 – much higher than any prediction from geomagnetic field models. Rates of change associated with the four spikes are ~0.35–0.55 μT/year (~0.7–1.1 ZAm2/year), at least twice the maximum rate inferred from direct observations spanning the past 190 years. The increase from 1750 BCE to 1030 BCE (73 to 161 ZAm2) depicts the Holocene’s largest change in field intensity.

Here we test the precursory enhancement in ionospheric total electron content (TEC) which has been reported by Heki (2011) and numerous Global Navigation Satellite System (GNSS) TEC observational studies before the 2011 Mw9.0 Tohoku-Oki and many great earthquakes. We verify the frequency of this TEC enhancement via analysis of a two-month vertical TEC (VTEC) time series that includes the Tohoku-Oki Earthquake using the procedure, based on Akaike’s information criterion, and threshold of Heki and Enomoto (2015). The averaged occurrence rate of the TEC enhancement is much larger than that reported by Heki and Enomoto (2015) when all of the visible GPS satellites at a given station are taken into account. We cannot rule out the possibility that the pre-seismic VTEC changes before the great earthquakes that were reported by Heki and Enomoto (2015) are not precursors but just a product of chance. We also analyze the spatial distribution of the pre-seismic TEC enhancement and co-seismic TEC depletion for the Tohoku-Oki Earthquake with the data after reducing inter-trace biases. We observe significant post-seismic depletion that lasted at least 2 h after the earthquake and extended at least 500 km around the center of the large-slip area. This means that evaluation of the enhancements using reference curves which was adopted by Heki 2011 and even by the recent papers (e.g. He and Heki 2016, 2017, 2018) is in danger of mistaking a large and long-lasting post-seismic TEC depletion for a pre-seismic enhancement.

Supervised deep learning models have become a popular choice for seismic phase arrival detection. However, they don’t always perform well on out-of-distribution data and require large training sets to aid generalization and prevent overfitting. This can present issues when using these models in new monitoring settings. In this work, we develop a deep learning model for automating phase arrival detection at Nabro volcano using a limited amount of training data (2498 event waveforms recorded over 35 days) through a process known as transfer learning. We use the feature extraction layers of an existing, extensively-trained seismic phase picking model to form the base of a new all-convolutional model, which we call U-GPD. We demonstrate that transfer learning reduces overfitting and model error relative to training the same model from scratch, particularly for small training sets (e.g., 500 waveforms). The new U-GPD model achieves greater classification accuracy and smaller arrival time residuals than off-the-shelf applications of two existing, extensively-trained baseline models for a test set of 800 event and noise waveforms from Nabro volcano. When applied to 14 months of continuous Nabro data, the new U-GPD model detects 31,387 events with at least four P-wave arrivals and one S-wave arrival, which is more than the original base model (26,808 events) and our existing manual catalogue (2,926 events), with smaller location errors. The new model is also more efficient when applied as a sliding window, processing 14 months of data from 7 stations in less than 4 hours on a single GPU.

Increasing ice flux from glaciers retreating over deepening bed topography has been implicated in the recent acceleration of mass loss from the Greenland and Antarctic ice sheets. We show in observations that some glaciers have remained at peaks in bed topography without retreating despite enduring significant changes in climate. Observations also indicate that some glaciers which persist at bed peaks undergo sudden retreat years or decades after the onset of local ocean or atmospheric warming. Using model simulations, we show that glacier persistence may lead to two very different futures: one where glaciers persist at bed peaks indefinitely, and another where glaciers retreat from the bed peak suddenly without a concurrent climate forcing. However, it is difficult to distinguish which of these two futures will occur from current observations. We conclude that inferring glacier stability from observations of persistence obscures our true commitment to future sea-level rise under climate change.